To investigate trends in aircraft noise exposure by airport and exposed population, we used noise exposure contours for U.S. airports, identified airport characteristics, and estimated exposed populations by race and ethnicity.
Noise evaluation
We obtained noise exposure contours (Figure 1) for 90 U.S. airports from the U.S. Department of Transportation (DOT) Volpe National Transportation Systems Center (Volpe). The airports included in this study constitute 18% of the Part 139 Federal Aviation Administration (FAA) certified airports, but accounted for 88% of the total number of in-flight aircraft in 2015. [26]. Noise contours were modeled for the years 1995, 2000, 2005, 2010, and 2015. Detailed information on the generation of aircraft noise contours is provided elsewhere. [24, 27]. Briefly, noise contours are created using the FAA’s Aviation Environmental Design Tool (AEDT), which was developed using internationally recognized methodologies for estimating the environmental impacts of aviation. Ta. Estimates are based on data from the 2000-2015 Enhanced Traffic Management System (ETMS) and the 1995 Official Aviation Guide (OAG), including airport runway locations and usage, as well as aircraft noise and performance (ANP) data. Formulated using standard aircraft profile data. database.
Figure 1
This study included a sample of U.S. airports (n = 90).
In this study, we used two noise metrics: DNL and Lnight. DNL reflects an average 24-hour noise exposure per year; add 10 dB(A) to measurements between 22:00 and 07:00, when noise sensitivity may be higher due to lower ambient noise. , which artificially imposes a penalty on nighttime hours. Nighttime is aggregation of nighttime noise exposure. DNL and Lnight were modeled in the range 45 to 75 dB(A) in 1 dB(A) steps.
We focused on three noise thresholds: (1) DNL 65, (2) DNL 45, and (3) nighttime 45 dB(A) level. The first one relates to the US regulatory threshold for significant noise exposure, and the latter two correspond to the World Health Organization recommended guidelines for aircraft noise exposure in the European region. [5]. The nighttime threshold is limited to 45 dB(A), which is the lowest modeling data available.
To exclude uninhabitable areas from the assessment, the contour lines were overlaid with a national regional waterway (i.e. water body) and green space geodatabase. A national regional hydrographic geodatabase (ponds, lakes, oceans, marshes, glaciers, rivers, streams, and canals) was available from the U.S. Census Topologically Integrated Geographic Encoding and Reference (TIGER) database. 2013 waterway database [28] The noise contours from 1995 to 2010 and the noise contours from 2016 were superimposed. [29] With contour lines for 2015. The 2010 National Green Spaces layer (Parks, Gardens, and Forests) was available from Esri. [30] Display all contour years overlapping.
Airport characteristics
Different airport characteristics by four U.S. Census regions (Midwest, Northeast, South, and West) and FAA hub type designations from 2001 (Passenger/Cargo Airline Hub Type, and Cargo Hub) has been identified. The FAA classifies major commercial airports (10,000 or more passengers annually) as hubs pursuant to 49 U.S.C. § 47102, and large hubs receive 1% or more of the U.S.’s annual commercial aircraft. , medium-sized hubs receive 0.25-1%, and small hubs receive 0.25-1%.0.05-0.25% for hub airports, less than 0.05% for non-hub airports but with more than 10,000 passengers per year [31]. Passenger/cargo airline hub types were categorized according to the airport’s primary passenger airline and cargo airline designation. Centralized LTO operations use a hub-and-spoke model, where an airline centralizes regional operations at a major central hub, or a point-to-point model, which is a direct AB operation without the need to go through a central hub. [32]. An airport is primary if it serves as a main central hub for a hub-and-spoke airline, secondary if it serves as a support hub for a hub-and-spoke airline, and a focal city if it is designated as a central airport for a point-to-point airport. was specified. – A point airline, or a non-hub/focus city if it does not function as a hub or focus city. The FAA classified an airport as a cargo hub if it ranked in the top 25 for total cargo landing weight. Airport passenger aircraft and cargo data were available from the Bureau of Transportation Statistics for 1995 and from the FAA Air Carrier Operations Information System (ACAIS) database for 2000–2015. [33, 34]. LTO operational data were available from the Air Traffic Activity System (ATADS) database from 1995 to 2015. [35].
Trend analysis of airport characteristics
First, we used response profile analysis across all 90 airports to estimate the average change in contour area over time. Analysis of the response profile allows the characterization of the pattern of change in the mean contour area over time. This method is suitable for longitudinal studies with balanced designs, when the timing of repeated measures is uniform across subjects, and for data that violate independence and homogeneity of variance assumptions. [36]. Contour area data were assumed to be complete for all 90 airports across each study time point and correlated across years for each airport. The covariance structure was selected by examining the nested model fit statistics table and likelihood ratio tests.
By identifying distinct groups of airports with common characteristics, rather than relying solely on fixed a priori factors, we are able to explore different characteristics of airports when investigating the association between aircraft noise exposure and health. could provide an informed approach to epidemiological studies to exploit We assessed inter-airport variability by statistically arranging airports by similarity using group-based trajectory modeling (GBTM). GBTM is a special application of finite mixture models that identifies distinct groups that share underlying properties and trajectories. [37]. We applied the SAS package Proc Traj using beta regression, which is appropriate for non-normal distributions. [38, 39]. The beta distribution dictates that the minimum and maximum area values for each year be used to normalize the noise contour areas to fall within the range 0 to 1. [40]. Model parameters were estimated using the maximum likelihood method. To determine the optimal number of groups, we started with a one-group model and gradually increased the number of groups.The best-fitting model was selected using the following criteria: log Bayes factor (2Δ BIC), Jeffries evidence scale for Bayes factor, nonoverlapping confidence intervals, posterior probability of group membership greater than 0.7, and Approaching a sufficient sample, ideally each group size is 5% or more [41, 42]. The BIC values were used to simultaneously determine the shape of each trajectory over time (i.e., the order of the polynomial relationship). Due to small cell sizes, Fisher’s exact test was used to test for nonrandom associations between traits and trajectory groups.
Trend analysis of exposed population
We assessed changes in the overall exposed population by Hispanic/Latino ethnicity and race as defined by the U.S. Census. Using U.S. Census designations, Hispanic/Latino ethnicity was categorized as people who self-identified as Hispanic or Latino and non-Hispanic/non-Latino. Race was categorized as White only, Black or African American only, Asian only, American Indian/Alaska Native, Native Hawaiian/Other Pacific Islander only, other race only, or two or more races. it was done. Population data were obtained from GeoLytics Inc. at the census tract level for 2000, 2005, 2010, and 2015. Since 2001, the decennial census racial classification has excluded “other races” and redistributed portions of “other races” and “two races.” or more races” to the remaining races. To maintain consistency in racial categories over time, we used race counts from the decennial census in 2000 and 2010, and the quinquennial American Community Survey (in 2005 and 2015). ACS) estimates were used. All census and ACS data were aligned at 2010 census tract boundaries. Comparative analysis from 2000 to 2015. Data for 2000 and 2015 were available from GeoLytics pre-weighted to the 2010 boundary, while 2005 data was obtained using geographic crosswalks available from the IPUMS National Historical Geographic Information System. interpolated to 2010 boundaries. [43]. We used simple area weighting to estimate the number of people living in noise-exposed areas. This is the sum of the percentage of masked noise contour areas that overlap the area multiplied by the estimated population within the overlapping area.
Exposed population estimates were evaluated in the following ways: (1) normalized by each subpopulation of the region; (2) By absolute numbers. (3) Normalized by the total population of the area. Areas were selected if they intersected the highest noise exposure contour (DNL 45, dB) during the study period (n = 13,416).[A] 2000). We defined these areas as “living near an airport.” We normalized by regional subpopulations to assess whether racial/ethnic groups had a disproportionate exposure burden (e.g., normalized by the total Hispanic/Latino population living within the region surrounding the airport). (Hispanic/Latino population), normalized by: Calculate the total population of a region to account for overall changes in population growth/decrease.
Analysis of trends in the exposed population by airport characteristics
To determine whether the socio-demographic characteristics of the population exposed to aircraft noise differ by airport characteristics over time, the number and normalized proportion of the exposed population when stratified by trajectory group We also investigated changes in We believe that this analysis may provide insight into the associations between common underlying characteristics that determine the trajectory of aircraft noise exposure and demographic characteristics such as ethnicity and race of the exposed population. I hypothesized that there might be.
Spatial analysis was completed using a common projected coordinate system within a geographic information system (GIS, Esri ArcGIS® Pro V2.2.3, Redlands, CA). Geographic areas were estimated in square kilometers (km2) after masking. Statistical analyzes were performed using Statistical Analysis System (SAS) v9.4 (Cary, NC).
